You are on page 1of 4

Virtual Anchors for Dummies M Blackman Jan 09

The virtual anchor concept is raised in Pengs 1978 paper and refers to a buried length at which a pipe's longitudinal expansion is totally suppressed by soil friction, ie the pipe is fully restrained. The length of piping required to form the virtual anchor is known as the active length. The axial forces pushing against soil restraint are due to pressure and thermal expansion. It helps to recognise that the force required to fully restrain a pipe does not change with its length, because a given temperature rise causes the same percentage strain. In Peng's paper a given length of soil provides a constant frictional force, whose value depends on burial depth and pipe weight, together with soil density and coefficient of friction with steel. Therefore, starting from a free end, the total restraint exerted by the soil on the pipe gradually increases until it reaches the fully restrained load at the virtual anchor. At locations further in from the virtual anchor, the pipe has zero expansion as it is fully held in place.. At the free end of the pipe, the total axial expansion is half of that which would otherwise be calculated for an aboveground pipe with an anchor at the virtual anchor location. This is calculated by taking the average of full axial restraint at one end and no restraint at the other. Soils with lower friction resistance or pipes with less depth of cover have longer active lengths and thus have greater expansion at the free end. In reality most pipes do not have a totally free end but have some resistance due to soil restraint as the pipe exits the ground and from the connection to above ground piping. This acts to reduce the expansion at the 'free' end. The model become more complicated when you consider that soil 'friction' develops over a finite distance rather than at zero movement - ref. American Lifelines Alliance Guidelines for Design of Buried Steel Pipe (GDBSP). Besides giving the problem a non-linear twist, this implies that active lengths should actually be longer and free end expansions larger.

Modelling in CAESAR II
Soil restraint is normally modelled as a series of bi-linear springs with nearzero stiffnesses above a certain yield point. If using the CII buried pipe modeller, the 'yield displacement factor' is provided in order for CAESAR to calculate restraint stiffnesses. This factor does dual duties, covering both axial and lateral restraints. The default figure provided by CAESAR (0.015) is appropriate for lateral stiffnesses (ref. Peng paper Part 2) but less appropriate for the axial direction. According to the GDBSP Appendix B, the soil displacement required to build up maximum axial resistance varies from 3mm for dense sand to 10mm for soft clay. Typical axial stiffnesses generated using the default yield displacement factor are much too low, giving yield displacements several

times too high for dense sand. If these figures are used, the actual free-end displacement and active length may be significantly larger than would be predicted by Pengs paper, but this would only be captured if sufficient length of pipeline were modelled. If a length of only one or two times theoretical active length is modelled, the growth and force would be underpredicted. If however an actual anchor was subsequently placed at a distance greater than the theoretical active length, the displacement would be overpredicted. My preferred approach is therefore to model a length of buried pipe of twice the Peng active length, then use the buried pipe modeller with default yield displacement factor (in order to reasonably estimate stiffnesses in lateral directions). Then take the model generated by CAESAR and manually adjust axial friction stiffnesses to approximate those implied by the GDBSP, and finally place an actual anchor at the far end of the line. The CAESAR model of course needs to have Bourdon effect aka pressure elongation switched on in the configuration settings. This makes up a significant portion of the movement. Why twice the theoretical active length ? To account for non-linear soil stiffness, which implies that the axial load increases asymptotically towards full restraint. A slightly greater free-end expansion is thus captured. If we didnt place an actual anchor at the far end we would need to increase this again to three or four times active length, otherwise there is not enough pipe on the other side of the virtual anchor to push back and hold it in place. The free end movement would not be overpredicted by adding more length. How can we tell if we've modelled sufficient length of buried pipe ? By checking the axial load along the pipe using the local force report and verifying that it increases up to around the fully restrained load as per the Peng method. If enough length has been modelled the axial load in the pipe will approximate the fully restrained load over one or more nodes, each with approximately zero movement. The fully restrained load in CAESAR would be slightly different to the Peng calculation due to simplification of formulae for longitudinal pressure stress. If we wanted to model the soil restraints in keeping with the Peng paper, this would imply using very high axial stiffnesses. The problem with this approach however is that the model will often fail to converge. Another approach we could take is when we are only interested in the effect of the buried pipeline movement on above ground piping and we don't want to use the buried pipe modeller. In this case we would model a length of pipe equal to half the active length plus some safety factor for non-linear soil friction, place an anchor at the far end and apply bourdon effect. Or we could just impose the estimated end displacement onto the above ground model.

Soil Restraint Comparison: CAESAR II 5.10 Method vs. ALA Guidelines Lateral Restraint CAESAR II buried pipe modeller is based on Peng formula multiplied by userdefined Overburden Compaction Multiplier (OCM) which defaults to 8. Ftr = OCM.0.5.g.((H+D/2).tan(45+/2))2 Where Ftr = transverse ultimate load per unit length of pipe H = depth to centre of pipe D = outer diameter of pipe = density of soil g = gravity acceleration = 9.81m/s2 = angle of internal friction of soil ALA Guideline figures are as follows Ftr = Nch.c.D + Nqh..g.H.D For cohesionless soils such as sand, the first term becomes zero. Nqh is an empirical factor which is tabulated against angle of internal friction and H/D ratio. Comparison of Ultimate Loads The following table uses an OCM of 1, ie the original Peng formula. Density is 1700 kg/m3. H=1.2, D=0.66 H=2.5, D=0.66 = 30 = 35 = 30 = 35 Nqh 6.5 10 8.2 12.8 Ftr (ALA) 86 133 225 351 Ftr (Peng) 58 72 200 246 Ratio 1.47 1.85 1.12 1.43 Table 1 Transverse Ultimate Load (kN per m of pipe length) H=3, D=0.66 = 30 = 35 8.7 13.6 287 450 277 341 1.03 1.32

If the ALA Guideline is considered to be current best practice then the OCM may be selected based on the ratios in Table 1, ie an OCM of 2 or less is appropriate for the parameters considered. Lateral Soil Stiffness Peng recommends using 1.5% of the depth to the bottom of pipe for the yield displacement, ie. Stiffness = Ftr/(0.015(H+D/2)). CAESAR II defaults to this value. ALA guideline suggests that the displacement at ultimate load is 4% of depth to bottom of pipe, however does not give an explicit formula for stiffness. Based on the example (non-linear) force-displacement curves in both ALA and Peng, the 1.5% figure is seen as appropriate. The effect of an increase in lateral soil stiffness is to increase bending stresses at branch connections but to reduce displacement and bending stresses at bends.

Axial Restraint CAESAR II figures are based on Peng : Fax = (2gD(H-D/2) + Wp) Where Fax = maximum axial friction load per unit length of pipe Wp = weight of pipe and content = coefficient of friction between pipe and soil ALA Guideline is as follows Fax = Dc + DHg.(1+K0)/2.tan(f.) For cohesionless soils again the first term becomes zero. The term tan(f.) is essentially equivalent to above, with f =0.7 for smooth steel. K0 is the coefficient of earth pressure at rest and is equal to 1-sin . Thus the equation can be written for sand as Fax = tan(f.).gDH(1- sin ) Comparison of Ultimate Loads The following table uses soil density = 1700 kg/m3, D=0.66, t=14.6 and =0.5. H=1.2m H=2.5m = 30 = 35 = 30 = 35 Fax (ALA) 11.9 13.5 24.9 28.1 Fax (Peng) 10.7 10.7 25.0 25.0 Ratio 1.1 1.3 1.0 1.1 Table 2 Axial Ultimate Load (kN per m of pipe length) Axial Soil Stiffness Peng does not address axial stiffness. Soil is considered infinitely stiff up until its ultimate load. ALA suggests a yield displacement of 3 to 5mm for sand in the axial direction. CAESAR II uses the same yield displacement as that calculated for the lateral direction, which is typically much higher than 5mm. For the range of parameters in Table 2, ALA calculated stiffnesses may be around 5 to 16 times those generated by the CAESAR II modeller. The main implications of using lower than actual axial soil stiffness are as follows More length to build up virtual anchor. Anchor blocks if used do not show full restraint unless significantly greater length of underground pipe is modelled. More expansion at free end of line and higher stresses at bends (but only if enough length is modelled or a physical anchor is modelled at theoretical virtual anchor location)

You might also like